An optimized multiobjective CPU job scheduling using evolutionary algorithms

Abstract

Scheduling in a multiprocessor parallel computing environment is an NP-hard optimization problem. The main objective of this work is to obtain a schedule in a distributed computing system (DCS) environment that minimizes the makespan and maximizes the throughput. We study the use of two of the evolutionary swarm optimization techniques, the firefly algorithm and the artificial bee colony (ABC) algorithm, to optimize the scheduling in a DCS. We also enhance the traditional ABC algorithm by merging the genetic algorithm techniques of crossover and mutation with the employed bee phase and the onlooker phase, respectively. The resulting enhanced ABC algorithm is used as the scheduling algorithm and is evaluated against the firefly and ABC algorithms. The results obtained show that in a distributed environment with a large number of jobs and resources, multiobjective scheduling using evolutionary algorithms can perform well in terms of minimizing makespan and maximizing throughput.

Low-latency and energy-efficient scheduling in fog-based IoT applications

“…Metaheuristic methods are used for resource allocation in parallel computing [35] and distributed computing [36]. The authors in [37] studied the task scheduling based on the metaheuristics approach in clouds.…”